2003-05-10 20:38 felicity
* Changes: Updated for 2.54rc1
2003-05-10 18:05 duncf
* lib/Mail/SpamAssassin.pm: Ready to release?
2003-05-10 17:34 felicity
* rules/: 20_head_tests.cf, 30_text_de.cf, 30_text_fr.cf,
30_text_it.cf, 50_scores.cf, 70_broken_rules.cf: bug 1891:
RCVD_IN_VISI is no longer available
2003-05-09 17:46 felicity
* lib/Mail/SpamAssassin/CmdLearn.pm: bug 1887: sa-learn didn't
accept parameters to -D
2003-05-09 17:28 felicity
* spamd/spamd.raw: bug 1818: spamd fails to create ~/.spamassassin
when running setuid with sql
2003-05-08 20:59 jmason
* spamd/libspamc.c: bug 1879: libspamc was closing fds it didn't
open
2003-05-08 16:28 jmason
* spamd/: README.spamd, spamc.pod: bug 1799: doco updated to note
protocol change compared to 2.4x, 2.6x
2003-05-08 14:47 felicity
* lib/Mail/SpamAssassin/BayesStore.pm: bug 1869: came up with a
kluge for the "tokens have 16-bit atime" problem. if the expire
code detects that scancount is > 65535, reset all the counters to
an atime of 0. it means one expire run doesn't actually expire
anything, but it should keep the useful tokens around instead of
the semi-random expire that was occuring in earlier versions.
2003-05-08 14:42 jmason
* spamd/libspamc.c: bug 1762: sscanf should not use %f due to
locale issues
2003-05-07 20:41 jmason
* spamd/spamd.raw: bug 1802: REPORT_IFSPAM is unhandled if message
is ham
2003-05-07 16:50 felicity
* rules/: STATISTICS-set1.txt, STATISTICS-set2.txt,
STATISTICS-set3.txt, STATISTICS.txt: bug 1589: generated
STATISTICS* files for the new GA run
2003-05-07 15:23 felicity
* lib/Mail/SpamAssassin/PerMsgStatus.pm: bug 1868:
X-Spam-Checker-Version may or may not have parens depending on if
the message was considered spam or not. Just add in the parens...
2003-05-07 13:06 jmason
* spamd/spamd.raw: bug 1725: spamd taint mode problem with ARGV
fixed
2003-05-07 05:38 felicity
* lib/Mail/SpamAssassin/BayesStore.pm: bug 1782: a potential but in
stdio could cause journal data to be corrupted by multiple writes
to the same area of the journal. switch to using syswrite()
instead of just print (stdio) to help address this issue.
2003-05-06 23:26 jmason
* lib/Mail/SpamAssassin/Bayes.pm: bug 1860: Bayes was not using
hapaxes
2003-05-06 06:14 felicity
* rules/: 20_meta_tests.cf, 50_scores.cf: bug 1589: spammers have
been targeting our nice rules to get themselves negative overall
scores. ran the GA again to lower the nice rule scores. this
should help deal with the forgeries. also added a "TOO_MANY_MUA"
rule that will catch when multiple USER_AGENT rules hit.
2003-05-01 23:25 jmason
* spamd/spamd.raw: bug 1799: REPORT should use local report
template, not default
2003-05-01 23:19 jmason
* sa-learn.raw, spamassassin.raw: bug 1737: Bayes doco improved
2003-04-23 21:48 felicity
* t/spamd_utf8.t: bug 1708: UTF-8 locale was specified incorrectly
(glibc version)
2003-04-22 05:14 felicity
* USAGE: bug 1739: USAGE file says that a test message will have
the subject rewritten, but rewrite_subject is off by default in
2.5x. change the docs to reflect this.
2003-04-22 05:00 felicity
* lib/Mail/SpamAssassin/BayesStore.pm: bug 1706: fixed a small
issue where the code was expecting a shift but got a pop, so we
need to look at the end of the array, not the beginning. :(
2003-04-22 04:54 felicity
* lib/Mail/SpamAssassin/BayesStore.pm: bug 1706: when bayes was
expiring tokens, it would check to see if there were enough tokens
left at the end and reprieve tokens if necessary. however, the
reprieval was happening in a random order, so the oldest tokens
would likely be left in the DB. the code will now sort the tokens
by atime and reprieval the most recently used tokens first.
2003-04-19 12:35 felicity
* lib/Mail/SpamAssassin/PerMsgStatus.pm: bug 1800: messages with
large positive scores (if they hit GTUBE at 1000 points for
instance), can have a X-Spam-Level header with more than the RFC
allowed number of chars. so we now limit X-Spam-Level to show a
max score of 100. that makes it obvious the message is really spam
while staying RFC compliant.
2003-04-10 08:27 felicity
* lib/Mail/SpamAssassin/EvalTests.pm: bug 1524: EvalTests looking
for/storing IPs would sometimes catch extra chars that occur around
the IP address (such as [ in Received headers). Use a new regexp
to only match IP addresses.
2003-04-06 18:29 felicity
* spamd/utils.c: bug 1755: a c++ comment slipped in, and some
compilers don't know what to do with it.
2003-04-06 12:56 felicity
* lib/Mail/SpamAssassin/Util.pm: bug 1741: allow ~ and space in
file path (good for Windows), return blank if blank passed in,
fixed return only if RE matches
2003-04-05 11:36 felicity
* lib/Mail/SpamAssassin/CmdLearn.pm: bug 1691: sa-learn was sending
non-error output to stderr instead of stdout (specifically the
"Learned # messages" string)
2003-04-03 17:48 duncf
* debian/: changelog.in, control, logcheck, spamassassin.default,
spamassassin.templates, watch, po/POTFILES.in, po/pt_BR.po,
po/templates.pot: Updated debian packaging for CVS builds; adds
po-debconf support
2003-04-03 17:48 felicity
* spamassassin.spec: updated spec file since 2.51 prepared for 2.54
release
2003-04-03 17:34 jmason
* build/preprocessor.pm: backport: preprocessor being indexed by
CPAN, avoid for next version
2003-04-03 17:33 jmason
* lib/Mail/SpamAssassin.pm: Starting 2.54
2003-03-31 17:42 jmason
* Changes, build/README, lib/Mail/SpamAssassin/BayesStore.pm: ready
for release
2003-03-31 16:43 jmason
* lib/Mail/SpamAssassin/BayesStore.pm: bug 1689: exists method does
not exist in NDBM_File for some reason. worked around
2003-03-31 16:08 jmason
* lib/Mail/SpamAssassin/Conf.pm: moved dialup_codes into more
privileged area
2003-03-29 19:26 jmason
* lib/Mail/SpamAssassin.pm: ready to release
2003-03-29 17:42 felicity
* MANIFEST, Makefile.PL, t/SATest.pm, t/rule_tests.t,
t/zz_cleanup.t: bug 1703: Makefile.PL would set a razor1 test if
razor1 wasn't installed. rule_tests.t was still trying to use
bytes.pm. spamd tests were using user's preferences/bayes/etc.
2003-03-29 17:28 felicity
* lib/Mail/SpamAssassin/BayesStore.pm: bug 1707: When syncing the
journal, the code took advantage of a UNIX-ism where you can open a
file, then rename it, and still access the file data.
Unfortunately, Windows doesn't let you do that. So we now rename
before we open, and add a "can we read the file" check beforehand
to avoid messy cleanup.
2003-03-29 17:10 felicity
* lib/Mail/SpamAssassin/Bayes.pm: bug 1699: a slight change to the
last patch, during tokenization, strip the extra chars before
skipping magic tokens.
2003-03-28 13:28 jmason
* MANIFEST, spamd/libspamc.c, spamd/libspamc.h, t/spamc_B.t,
t/data/spam/bsmtp: bug 1595: BSMTP mode (spamc -B) fixed
2003-03-28 12:26 felicity
* lib/Mail/SpamAssassin/: Bayes.pm, BayesStore.pm: bug 1699:
slightly better regex. ignore actual magic token strings, not just
magic token prefixes. this leaves more tokens able to be used.
2003-03-28 12:14 felicity
* lib/Mail/SpamAssassin/: Bayes.pm, BayesStore.pm: Bug 1699: The
code wasn't protecting the Bayes DB "magic" tokens (tracks num
ham/spam, etc.) from being overwritten if they occured in a
message. This allows mails to corrupt the DB either maliciously or
if they discuss the tokens.
2003-03-27 22:41 jmason
* lib/Mail/: SpamAssassin.pm, SpamAssassin/BayesStore.pm,
SpamAssassin/CmdLearn.pm, SpamAssassin/DBBasedAddrList.pm,
SpamAssassin/Locker.pm, SpamAssassin/UnixLocker.pm,
SpamAssassin/Win32Locker.pm: bug 1690: now has different lock
timeouts for different bayes operations
2003-03-27 19:25 felicity
* lib/Mail/SpamAssassin/Bayes.pm: bug 1696: you could continuously
"forget" tokens from messages you've learned, and even those you
hadn't learned.
2003-03-27 19:18 felicity
* spamassassin.raw, t/reportheader.t: bug 1695: report_header was
removed in 2.5x but still was in the POD docs and a test config.
2003-03-27 19:16 felicity
* sa-learn.raw: bug 1502: sa-learn wasn't paying attention to the
compile-time paths for rules and the like. The problem was that
the variables were declared my() which aren't accessible by other
packages.
2003-03-27 17:27 jmason
* lib/Mail/: SpamAssassin.pm, SpamAssassin/BayesStore.pm,
SpamAssassin/CmdLearn.pm, SpamAssassin/Conf.pm,
SpamAssassin/DBBasedAddrList.pm, SpamAssassin/Locker.pm,
SpamAssassin/UnixLocker.pm, SpamAssassin/Win32Locker.pm: backing
out lock_timeout patch for redo
2003-03-27 17:01 jmason
* lib/Mail/: SpamAssassin.pm, SpamAssassin/BayesStore.pm,
SpamAssassin/CmdLearn.pm, SpamAssassin/Conf.pm,
SpamAssassin/DBBasedAddrList.pm, SpamAssassin/Locker.pm,
SpamAssassin/UnixLocker.pm, SpamAssassin/Win32Locker.pm: bug 1690:
now has different lock timeouts for different bayes operations
2003-03-27 16:27 jmason
* spamassassin.raw: bug 1608: spamassassin.1 lists incorrect paths
for config files
2003-03-27 16:15 jmason
* spamd/: libspamc.c, spamc.c, spamc.pod: bug 1697: spamc fails to
dump message when running out of memory
2003-03-27 13:18 jmason
* spamd/spamd.raw: bug 1536: kernel reporting SIGWAIT problem on
recent Red Hat kernels
2003-03-27 13:13 jmason
* lib/Mail/SpamAssassin/DBBasedAddrList.pm: bug 1661: user_prefs
was world-writable after AWL tie, due to failure to set umask
correctly
2003-03-27 13:12 jmason
* lib/Mail/SpamAssassin/EvalTests.pm: bug 1401: check_rbl was
checking incorrect number of Received hdrs
2003-03-27 12:35 felicity
* lib/Mail/SpamAssassin/: Dns.pm, Reporter.pm: bug 1632: Razor
modifies %ENV so we tried to use a local() to make a temp version,
which works, but kills %ENV for the Razor code as well. So now
we'll let the helper_mode functions handle backing up/restoring
%ENV as appropriate.
2003-03-27 09:55 felicity
* spamd/spamd.raw: bug 1514: spamd was trying to use "../lib" which
is ok most of the time, but during a "make test" on perl 5.005 it
errored out with the "bytes.pm" error. so we now check for a
"../blib/lib" first and use it if it exists. otherwise, we
continue with "../lib".
2003-03-27 09:48 felicity
* lib/Mail/SpamAssassin.pm: bug 1685: we should clean PATH when we
start if we're in taint mode. This fixes a Sys::Hostname problem
in 5.005, but it's a good idea anyway since we won't be able to use
PATH unless it's cleaned in taint mode anyway.
2003-03-26 17:53 jmason
* lib/Mail/SpamAssassin.pm: bug 1630: add-all-addrs-to-blacklist
was a no-op, due to > vs >= bug
2003-03-26 12:18 jmason
* sa-learn.raw, lib/Mail/SpamAssassin/CmdLearn.pm: bug 1572:
sa-learn doco was still talking about auto-whitelists, although
this was never implemented
2003-03-26 12:08 jmason
* spamd/README.spamd: bug 1674: README.spamd didn't know that
auto-learning had been implemented
2003-03-26 12:06 jmason
* MANIFEST.SKIP: updated MANIFEST.SKIP likewise
2003-03-26 12:05 jmason
* spamd/.cvsignore: updated cvsignore to ignore sslspamc
2003-03-26 12:04 jmason
* spamd/spamd.raw: bug 1684: syslog-socket had a typo, fixed
2003-03-26 12:01 jmason
* spamd/libspamc.c, spamd/utils.c, spamd/utils.h, .cvsignore: bug
1637: compilation of spamd/sslspamc broken, fixed
2003-03-26 11:13 jmason
* config.h.in, configure, configure.in, .cvsignore, .cvsignore: bug
1542: configure doesn't detect SSL_CTX_free on NetBSD, fixed
2003-03-26 10:20 msquadrat
* configure, Makefile.PL, .cvsignore: bug 1542: configure wasn't
updated, so the fix for this bug didn't work per default
2003-03-24 17:22 jmason
* lib/Mail/SpamAssassin.pm: 2.53 devel cycle started
2003-03-24 17:20 jmason
* Changes: 2.52 RELEASED
2003-03-24 17:11 jmason
* INSTALL: last-minute doco change: recommend using DB_File
2003-03-24 14:39 msquadrat
* lib/Mail/SpamAssassin.pm: ready for release
2003-03-24 14:21 msquadrat
* spamassassin.raw: bug 1260: corrected description of
--nocreate-prefs option
2003-03-24 14:17 msquadrat
* spamd/netbsd-rc-script.sh: bug 1656: updated init script for
NetBSD
2003-03-24 13:28 msquadrat
* configure.in: bug 1542: configure didn't detect SSL_CTX_free on
NetBSD
2003-03-24 13:11 msquadrat
* spamd/spamd.raw: bug 1601: option --syslog-socket wasn't
implemented
2003-03-23 17:10 felicity
* lib/Mail/SpamAssassin/: Bayes.pm, BayesStore.pm: bug 1664: the
only entries in the bayes journal are token access time updates
(from scans). we can exploit the fact that most of the tokens
appear multiple times by caching in memory the updates per token,
then just writing to db the end result times for all tokens.
also, scan() tries to opportunistically do a journal sync and
expire which causes lots of contention if there is a lot of
messages for a single db (this is usually seen with site-wide dbs).
so we flag when we are doing the sync and expire and let all other
pids not be so opportunistic. this should greatly lower
contention. we also let the db tie r/w happen before we try to
read from the journal, and also exploit the fact we can open the
journal before renaming it so the most common error occurs first,
and then a journal read failure means we don't need to rename the
temp file back.
2003-03-23 16:52 felicity
* lib/Mail/SpamAssassin/Bayes.pm: bug 1672: a typo in a backported
patch for 2.51 caused bayes to sometimes not unlock the db.
2003-03-20 12:31 jmason
* lib/Mail/SpamAssassin.pm: 2.51 devel cycle started
2003-03-20 12:27 jmason
* Changes, build/update_devel, lib/Mail/SpamAssassin.pm: 2.51
RELEASED
2003-03-20 12:07 jmason
* Changes, lib/Mail/SpamAssassin.pm: 2.5.1 RELEASED
2003-03-20 11:38 jmason
* lib/Mail/SpamAssassin.pm: ready to release
2003-03-19 22:53 jmason
* sa-learn.raw, lib/Mail/SpamAssassin/CmdLearn.pm: bug 1638:
--force-rebuild doco is unclear. fixed
2003-03-19 22:51 quinlan
* lib/Mail/SpamAssassin/Conf.pm: trivial documentation correction
2003-03-19 14:26 msquadrat
* Makefile.PL: bug 1652: Perl 5.6.0 and 5.6.1 ship different
versions of ExtUtils::MakeMaker; both carry the version 5.45 though
2003-03-19 11:29 felicity
* Makefile.PL: bug 1652: the suggested patch doesn't work. perl
5.6.0 and perl 5.6.1 both have EU:MM 5.45, but the one in 5.6.1 has
PM_FILTER. so reverting the initial patch since it doesn't take
this into account.
2003-03-18 21:27 felicity
* lib/Mail/SpamAssassin/Bayes.pm: bug 1655: we tie to check # of
Bayes tokens, but we don't untie if # ham/spam is too low. fixed.
2003-03-18 20:52 jmason
* MANIFEST: bug 1541: removed trim_bayes_db from distro
2003-03-18 18:38 jmason
* lib/Mail/SpamAssassin/BayesStore.pm: bug 1636: incorrect errors
from bayes if an error occurred during big scan-counter increment
2003-03-18 17:56 felicity
* lib/Mail/SpamAssassin/BayesStore.pm: bug 1626: if a token's atime
in the bayes db is > current scan count, there was an expiry error,
so we'll reset the atime to the last expire time.
2003-03-18 15:13 quinlan
* lib/Mail/: SpamAssassin.pm, SpamAssassin/Conf.pm,
SpamAssassin/PerMsgStatus.pm: bug 1576: need safer way to render
attachments for broken clients
2003-03-18 14:52 felicity
* t/: razor.t, razor2.t: bug 1619: razor tests were failing because
the rule description changed from what was being tested
2003-03-18 14:11 felicity
* lib/Mail/: SpamAssassin.pm, SpamAssassin/Bayes.pm,
SpamAssassin/BayesStore.pm, SpamAssassin/PerMsgStatus.pm: bug 1566:
if spamd is run with -x, spamd only ties the Bayes DBs once, so SA
doesn't detect when nspam/nham have gone over the min # required
(200) and enable automatically.
2003-03-18 14:04 felicity
* lib/Mail/SpamAssassin/BayesStore.pm: bug 1579: there were several
undef warnings in the Bayes code. cleaned that out.
2003-03-18 13:58 felicity
* lib/Mail/SpamAssassin/: Bayes.pm, BayesStore.pm, PerMsgStatus.pm:
bug 1556: PerMsgStatus wasn't cleaning up bayes DB ties after it
was done scanning which caused problems with MailScanner.
2003-03-18 13:55 felicity
* lib/Mail/SpamAssassin/HTML.pm: bug 1653: HTML tests had undef
warnings sometimes. do better parsing via RE.
2003-03-17 19:27 felicity
* spamd/spamd.raw: bug 1640: spamd was sending the Content-Length
header instead of the Spam: header for protocol 1.2 clients. it
should have been sending just the Spam header, not the
Content-Length one.
2003-03-17 19:21 felicity
* lib/Mail/SpamAssassin/Conf.pm: bug 1648: unwhitelist_from_rcvd
was non-functional due to typo. also updated some docs.
2003-03-17 19:14 felicity
* Makefile.PL: bug 1652: we checked for version < 5.45, but the
feature was added > 5.45, so we get confused when == 5.45.
2003-03-17 15:56 felicity
* rules/: 10_misc.cf, 20_anti_ratware.cf, 20_body_tests.cf,
20_compensate.cf, 20_head_tests.cf, 20_html_tests.cf,
20_meta_tests.cf, 20_phrases.cf, 20_porn.cf, 20_ratware.cf,
20_uri_tests.cf, 23_bayes.cf, 25_body_tests_es.cf,
25_body_tests_pl.cf, 25_head_tests_es.cf, 25_head_tests_pl.cf,
30_text_de.cf, 30_text_es.cf, 30_text_fr.cf, 30_text_it.cf,
30_text_pl.cf, 50_scores.cf, 60_whitelist.cf, 70_broken_rules.cf,
70_cvs_rules_under_test.cf, regression_tests.cf: bug 1607: add
license information to the rule files
2003-03-17 11:07 felicity
* lib/Mail/SpamAssassin/Bayes.pm: bug 1651: if for some reason a
Bayes scan() doesn't occur (too few tokens, tokens without enough
probability, etc,) but the DBs were already tied, we didn't untie
them.
2003-03-17 07:24 felicity
* sa-learn.raw, spamassassin.raw, spamd/spamc.pod: bug 1605: SA
documentation ought to specify what "ham" is for people who don't
necessarily know it means non-spam.
2003-03-17 07:19 felicity
* lib/Mail/: SpamAssassin.pm, SpamAssassin/Bayes.pm,
SpamAssassin/EvalTests.pm, SpamAssassin/PerMsgStatus.pm: fix for
bug 1591: Bayes is_available() didn't pay attention to use_bayes,
so it would still attempt a tie and possibly create DBs even if it
wasn't going to use the bayes stuff. renamed the function to
is_scan_available() since other modules call it to see if a scan
would be done (determining scoreset, etc.) cleaned up scan() a
little (let is_scan_available check for # of tokens)
2003-03-17 07:12 felicity
* lib/Mail/SpamAssassin/BayesStore.pm: fix for bug 1610: after a
bayes expire, the current message scan count would return to 0.
this is because the magic token in the db (scancount_base) wasn't
being copied over. oops. also added checks for return values to
make sure we don't delete the journal unless we really should.
modified expiry_due to do an expire if the last expire time is >
current message scan. this should never happen in actual usage and
only appears if this bug has occured.
2003-03-16 23:25 quinlan
* debian/: README.Upgrade, changelog.in, copyright,
spamassassin.README.Debian, spamassassin.docs,
spamassassin.examples: bug 1643: backport Debian 2.50 changes to
2.50 branch
2003-03-15 10:29 msquadrat
* lib/Mail/SpamAssassin/Dns.pm: bug 1641: don't load razor if it is
not available
2003-03-15 10:24 msquadrat
* t/: SATest.pm, db_awl_path.t, reportheader.t, whitelist_to.t: bug
1639: permission issue in testbench
2003-03-15 10:19 msquadrat
* lib/Mail/SpamAssassin/UnixLocker.pm: bug 1635: warn when removing
a stale lock file fails
2003-03-15 10:16 msquadrat
* lib/Mail/SpamAssassin/BayesStore.pm: bug 1625: problem with
sync_journal() when journal is out-of-date
2003-03-15 08:41 msquadrat
* lib/Mail/SpamAssassin/CmdLearn.pm: bug 1545: possible warning
about undefined veriable in sa-learn
2003-03-15 08:33 msquadrat
* sa-learn.raw, lib/Mail/SpamAssassin/CmdLearn.pm: bug 1578: fix
problem with --single in sa-learn
2003-03-15 08:23 msquadrat
* rules/30_text_de.cf: bug 1568: small type in German translation
2003-03-15 08:19 msquadrat
* sa-learn.raw, lib/Mail/SpamAssassin/BayesStore.pm: bug 1545:
extended docu for sa-learn
2003-03-15 08:10 msquadrat
* lib/Mail/SpamAssassin/: Bayes.pm, BayesStore.pm, UnixLocker.pm,
Win32Locker.pm: bug 1545: locking problems in combination with
GDBM_File
2003-03-15 08:03 msquadrat
* lib/Mail/SpamAssassin/BayesStore.pm: bug 1513: permission issues
with bayes_msgcount
2003-03-15 07:56 msquadrat
* lib/Mail/SpamAssassin/Conf.pm: bug 1528: wrong clear statement in
unsafe_report docu
2003-03-15 07:50 msquadrat
* lib/Mail/SpamAssassin/PerMsgStatus.pm: bug 1570: tab instead of
space in X-Spam-Status header
2003-03-14 10:49 quinlan
* lib/Mail/SpamAssassin/Conf.pm: bug 1554: documentation bug for
ok_locales
2003-03-14 10:48 quinlan
* lib/Mail/SpamAssassin/PerMsgStatus.pm: fix bug 1546: use our
local_tz() function instead of non-standard strftime %z extension
to get local timezone
2003-03-13 19:59 felicity
* lib/Mail/SpamAssassin/BayesStore.pm: fix for bug 1611
2003-03-13 19:34 felicity
* lib/Mail/SpamAssassin/NoMailAudit.pm: fix for bug 1544
2003-03-13 19:23 felicity
* spamd/spamd.raw: fix for bug 1604
2003-03-13 19:10 felicity
* lib/Mail/SpamAssassin/EncappedMessage.pm,
masses/rewrite-cf-with-new-scores, masses/score-ranges-from-freqs:
bug 1558
2003-03-13 18:59 felicity
* spamassassin.spec: bug 1593, and updated spec file to 2.51
version
2003-03-13 18:50 felicity
* lib/Mail/SpamAssassin/Util.pm: fix for bug 1559
2003-02-24 08:09 msquadrat
* lib/Mail/SpamAssassin/Conf.pm: backport fix for bug 1537: better
docu of dns_available